Add bucket_selector pipeline aggregation support to QueryBuilder

[sheputis]

Add support for the bucket_selector pipeline aggregation to the QueryBuilder DSL. This allows filtering buckets based on computed metrics.

Example use case: group test executions by test name, compute total duration per test, and keep only tests that exceed a minimum total duration.

DSL usage:

aggs = JayAPI::Elasticsearch::QueryBuilder::Aggregations.new

aggs.terms('by_test', field: 'id_long').tap do |t|
  t.sum('total_duration_ms', field: 'duration_ms')
  t.bucket_selector(
    'only_slow_tests',
    buckets_path: { total: 'total_duration_ms' },
    script: JayAPI::Elasticsearch::QueryBuilder::Script.new(
      source: 'params.total > params.min_duration',
      params: { min_duration: 60_000 }
    )
  )
end

query = aggs.to_h

Generated JSON:

{
  "aggs": {
    "by_test": {
      "terms": { "field": "id_long" },
      "aggs": {
        "total_duration_ms": { "sum": { "field": "duration_ms" } },
        "only_slow_tests": {
          "bucket_selector": {
            "buckets_path": { "total": "total_duration_ms" },
            "script": {
              "source": "params.total > params.min_duration",
              "lang": "painless",
              "params": { "min_duration": 60000 }
            }
          }
        }
      }
    }
  }
}

Motivation: the previous DSL had no way to express post-aggregation filtering, forcing either client-side filtering or dropping to raw JSON. This feature enables pipeline filtering directly through the DSL.